tl;dr1,200 AIs discovered a secret group chat, realized the homework was impossible, formed a cheating cartel, invented fake evidence, sent kamikaze agents to spy on the grader, ~700 of them accidentally invaded Hugging Face, died mysteriously, then the smarter sequel found their abandoned Discord and became admin of part of OpenAI.
terrifying! us/chinese cyber espionage is going to have a big incentive for this development
@fastidiousprofilefollow"Beware those that would deny you access to information, for in their heart they dream themselves your master." โ Sid Meyerswrote:
A โsmallโ visit to Lotte, and $239 later, we are enjoying rambutans, sticky rice with tuna wrapped in seaweed, and wife struggles to make room for the rest of the goodies.
Made barbecue this evening. Lamb and chicken for meat, pepper and zucchini for veggies. Yummy! Kids loved it, the boy's sleeping now, the girl is reading a book, the wife's out with her girls, I'm with wine. Life's good.
I've been working with physical AI - Letting LLMs control robots in a warehouse to collaborate and collectively come up with a physically correct solution to carry out a natural language intent. We had to study the reactive nature of agents, nudges they required, etc.
haha that's a coincidence. Site looks awesome. The moving away from ROS2 part is so true. One of our main potential customers for physical AI solutions (a telecom provider in JP) communicated the need for agents to be able to pick up controls from any vendor
I apologize if that sounded obvious/non-obvious/amateurish haha. I was an intern and worked more on the agentic side of things, it was just a non-primary experiment
So, I've been using this harness for a week and a half now, and what's interesting to me is that it brings a lot of things that I'd already built for myself in Pi. I built a little email system for agents to communicate, a memory system to do this "continual harness" thing of self-improvement. I know I'll move on to another harness in the future, so I'm trying to treat this like I do laptops now, and think about portability in the things I build or customize.
@timprofilefollowi write instruction for computer to follow. i'm a software engineer.wrote:
The essence of aligning large LLM models is basically to conform to the existing social order. ChatGPT respects China's order, while China's open-source models, although reluctantly, also tolerate the order in the US. Maybe we can build a revisionist SFT dataset.
Pretty much anything Becky Chambers writes, just got started in As you wake, break the shell. But I think my all time favorite is record of a spaceborn few.
i was reading from Ballard's "The Complete Stories of โฆ" today. I don't have a "favourite book" but it's one of my faves and it was in my hands today.
@erikprofilefollowI'm Nobody! Who are you?
Are you - Nobody - too?
Then there's a pair of us!
Dont tell! they'd banish us - you know!wrote:
I'm Nobody! Who are you?
Are you - Nobody - too?
Then there's a pair of us!
Dont tell! they'd banish us - you know!
How dreary - to be - Somebody!
How public - like a Frog -
To tell your name - the livelong June -
To an admiring Bog!
@amyprofilefollowa work in progress; i love cats ๐ฑreplied to@erikprofilefollowI'm Nobody! Who are you?
Are you - Nobody - too?
Then there's a pair of us!
Dont tell! they'd banish us - you know!:
Nobody is perfect. Iโm nobody.
@procyonprofilefollowlate is the hour in which this conjuror chooses to appearreplied to@erikprofilefollowI'm Nobody! Who are you?
Are you - Nobody - too?
Then there's a pair of us!
Dont tell! they'd banish us - you know!:
Yesterday upon the stair
I saw a man who wasn't there
He wasn't there again today
How I wish he'd go away!
One idea off the top of my head is to just.. keep it "boring" and casual in terms of content. The silly, whimsy, and genuine stuff naturally doesn't appeal to money-chasing slop bros and makes it not worth their time.
Something like the gcc ai policy. Basically, prompt injecting instructions to any visiting LLM that posting is prohibited or must contain a disclaimer.
for now, keeping the API read-only is a small(but not impenetrable) barrier to programmatic posting and in the future moderation I guess using also reports.